Discussion:
Silicon Motion 712 (Lynx EM+) and AMD Alchemy
(too old to reply)
Lewin A.R.W. Edwards
2004-04-03 19:10:05 UTC
Permalink
fine, except the Epson doesn't have enough memory bandwidth to drive both and
LCD and a CRT display at the same time which is a requirement of this
1) Can I just use the VESA frame buffer driver with this setup? (That is, VESA
claims to be x86 only, but will it work in this setup?)
The question is kind of bass-ackwards. The VESA framebuffer driver is
not really "x86-specific" as such, but the whole idea is predicated on
the existence of a VESA BIOS in your target system. An extended int
10h call is made to set the video mode. If your system doesn't have a
VESA BIOS extension (which it won't if it's not a vanilla x86 system)
then the concept of using the VESA driver is meaningless.
2) If not, does a frame buffer driver exist for the SM712 chip?
There is no direct kernel support for the Lynx series. Are you
planning to use this chip as a long-term solution? TTBOMK SMI is out
of that business (retail graphics chips) - at least they claimed to be
when we last tried to talk to them a couple of years ago. I would say
that Trident is a better choice for future production.

By the way, have you considered one of the other Epson parts, e.g.
SED1386, which can support CRT or LCD? Or do you need dual-head
operation with different outputs on CRT and LCD?
Ted Schroeder
2004-04-14 16:08:26 UTC
Permalink
Post by Lewin A.R.W. Edwards
fine, except the Epson doesn't have enough memory bandwidth to drive both and
LCD and a CRT display at the same time which is a requirement of this
1) Can I just use the VESA frame buffer driver with this setup? (That is, VESA
claims to be x86 only, but will it work in this setup?)
The question is kind of bass-ackwards. The VESA framebuffer driver is
not really "x86-specific" as such, but the whole idea is predicated on
the existence of a VESA BIOS in your target system. An extended int
10h call is made to set the video mode. If your system doesn't have a
VESA BIOS extension (which it won't if it's not a vanilla x86 system)
then the concept of using the VESA driver is meaningless.
Thanks for that tip. I know that X has some int10 simulation stuff in it. I'll
take a closer look at that.
Post by Lewin A.R.W. Edwards
2) If not, does a frame buffer driver exist for the SM712 chip?
There is no direct kernel support for the Lynx series. Are you
planning to use this chip as a long-term solution? TTBOMK SMI is out
of that business (retail graphics chips) - at least they claimed to be
when we last tried to talk to them a couple of years ago. I would say
that Trident is a better choice for future production.
Hmm. That's news to me. We're finding no problem getting SM712 chips in plenty
of volume for the forseeable future. I will ask our hardware guy about this.
Post by Lewin A.R.W. Edwards
By the way, have you considered one of the other Epson parts, e.g.
SED1386, which can support CRT or LCD? Or do you need dual-head
operation with different outputs on CRT and LCD?
We're using the SED1386 now and it, quite honestly, is not even close to
powerful enough to get the job done. It barely supports 800x600x16 and when you
put it in that mode the memory bandwidth is so limited that we had to reduce the
refresh rate to about 30 Hz to allow the CPU enough access to the VRAM so that
screen paints didn't take several seconds. This same memory bandwidth problem is
the reason why right now we can only output to either the CRT or the LCD. Not
enough bandwidth to make them both run at the same time. In addition, we're
going to 1024x768 size screens which aren't supported at all by the 1386.

Ted

Loading...